A stochastic model for the sea level in the Estonian coastal area

A stochastic model is suggested to perform the space–time optimal analysis of the sea level data recorded in 1978–1982 at 20 stations at the Estonian coast and along a part of the Latvian coast surrounding the Gulf of Riga. The original time series recorded with the time lag of 1, 6 or 12 h are divided into mean and fluctuation components. The mean field is modeled as the sum of the linear trend and annual harmonic. The mean sea level is generally higher at the stations located in the river mouth area. The estimated linear trend yielding the sea level rise of 1–3 cm/year is an approximation of the interannual variability over the selected 5-year period. The dominating annual harmonic with amplitude of 20 cm describes 40–45% of the total variability of the time series of the monthly mean sea level values. The temporal and spatial correlations of the sea level fluctuation field were estimated on the basis of the suggested stochastic model. The correlation functions were approximated by Gaussian functions yielding the temporal correlation radius (e-folding scale) of about 10 days and spatial correlation radius of 200–400 nm. According to the developed criterion, proceeding from the suggested stochastic model, at least 90% of the sea level data from the Estonian coastal area should be considered as meeting the quality requirements. There was no significant difference in the quality of data measured either continuously by mareographs or observed by reading the bench sticks. After removal of outliers, the approach was utilized to reconstruct the sea level field in the Estonian coastal area in 1978–1982 with an acceptable low reconstruction error.